You are on page 1of 39

Performance Evaluation of Dynamic

Systems

• Open-Loop System vs. Closed-Loop


System
• Indication of of Performance
• Performance of the First and Second
Order Systems
• Effects of Zeros
• Performance of General Dynamic Systems
Open-Loop System vs. Closed-Loop System
Let’s take a look at the armature-controlled motor system:

Normally, we want the input instruction signal to be the same kind as that of the
output (controlled) signal:
Ei(s)=Ea(s)
Kp G(s)

Potentiometer Motor
Open-Loop System vs. Closed-Loop System
Clearly the performance of this motor system can be described by:

Problems with the open-loop system: The performance is solely determined by


motor itself, i.e., G(s). Therefore, if motor design cannot yield satisfied dynamic
performance such as transient response time, stability, tracking error, etc., then
we essentially could do nothing with this set-up. But one cannot, so far, expect
that motor design would be conducted to address the issues mentioned here.
Instead the main issues addressed in motor design are related to size, body
vibration and balance, efficiency, etc.
Open-Loop System vs. Closed-Loop System
The desire to actively regulate performance of signal response motivates us to
use the closed-loop system structure.
Step1: Make available

Motor

Comparator(OP)
Ei(s)
+ Ea(s)
Kp G(s)
-

Em(s)
Ks
Potentiometer

Sensor( encoder, potentiometer, or rotary transformer)


Open-Loop System vs. Closed-Loop System
Step 2: Closing the loop with a to-be-designed element called Controller C(s)

Motor

Comparator(OP)
Ei(s) Ea(s)
+
Kp C(s) G(s)
-

Em(s)
Ks
Potentiometer

Sensor
Open-Loop System vs. Closed-Loop System
For the closed-loop motor system, we have

If Kp and Ks can be calibrated as Ks= Kp=K, then we have

Therefore, C(s) gives us a degree of freedom when we evaluate the


performance yielded by the transfer function.
Open-Loop System vs. Closed-Loop System
So the question here is to ask: if the transfer function

does not produce the desired performance, could we find a C(s) such that
the transfer function

would yield satisfied performance?


Either way, we have to learn how to evaluate the performance of a transfer
function

R(s) Y(s)
Transfer function
=Good or Bad?
Indication of Performance
In general, there are four kinds of performance that we are concerned about:

1. Stability:

R(s) Y(s)
G(s)=B(s)/A(s)

R(s) Y(s) R(s) Y(s)


1/(s+1) 1/(s-1)

r(t)=1(t), r(t)=1(t),
Indication of Performance
2. Transient performance: given r(t), how does y(t) catch r(t), fast or slow, any
oscillation, etc.?
Indication of Performance
3. Accuracy:

R(s) Y(s)
G(s)=B(s)/A(s)

If is zero, then there is no tracking error;

if is not zero, then the tracking will not perfect.

Obviously, for tracking accuracy, the closed-loop structure is essentially


superior to the open-loop structure.
Indication of Performance
4. Robustness: how does the performance such as stability, transient
performance or tracking accuracy vary when internal/external parameters
or environment change?

R(s) Y(s)
G(s)=B(s)/A(s)

Robust performance is normally reflected in the so-called sensitivity function


with respect to a specified factor, for example:

The sensitivity function S(s) about the parameter a can be obtained as


Indication of Performance
Testing signals:
Performance of the First and Second Order
Systems
Performance of the first-order system (a>0):

Unit-step response:
Given r(t)=1(t), we have

Time constant T=1/a, at which: c(T)=1-e-1=0.632


and
Note that if then c(t)=1(t), i.e., there will be no transient process.
Performance of the First and Second Order
Systems
Unit-ramp response:
Given r(t)=t, we have

and

Once again, if then c(t)=t, there will be no transient process and

=0

Sensitivity function about T:

In summary, the performance of the first order system can be characterized by


the time constant T
Performance of the First and Second Order
Systems
Performance of the second-order system:
R(s) C(s)
G(s)

where Damping ratio

Natural frequency

Note that

Consider the following three cases:

Under-damped Critically damped Over-damped


Performance of the First and Second Order
Systems
For under-damped case: and

Damped natural frequency

Unit-step response: given r(t)=1(t) or R(s)=1/s, then

Note that the time constant T=1/


Performance of the First and Second Order
Systems
Performance of the First and Second Order
Systems
For critically damped case: and

Unit-step response:

For over-damped case: and

Unit-step response:
Performance of the First and Second Order
Systems
Performance of the First and Second Order
Systems
Performance of the First and Second Order
Systems
Performance of the First and Second Order
Systems
Step responses of second-
order
under-damped systems as
poles
move:
a. with constant real part;
b. with constant imaginary part;
c. with constant damping ratio
Performance of the First and Second Order
Systems
It is the under-damped case that is of the most interesting and important:

Unit-step response of the second-order system


Performance of the First and Second Order
Systems
Rise time Tr : the time required for the response to go from 0.1 of the final value
to 0.9 of the final value
Performance of the First and Second Order
Systems
Peak Time Tp: the time required to reach the first, or maximum, peak.
It can be obtained from the following derivation:

Settling Time Ts: the time required for the transient’s damped oscillations to
reach and stay within 2% of the steady-state (final) value.

It can be estimated as: Ts=4T=4/


Performance of the First and Second Order
Systems
Percentage Overshoot Mp: Mp is defined as

And it can be calculated from (note that cfinal=1):


Performance of the First and Second Order
Systems
Geometric representation of performance indices:
Performance of the First and Second Order
Systems
Example 1:

Given the pole plot of the standard second order system, find all the
Performance indices defined.
Solution:
Performance of the First and Second Order
Systems
Example 2:
R(s) + C(s)
+
K/(Js+B) 1/s
- -

Kh

Given J=1kg-m2 and B=1 N-m, determine the values of K and Kh so that
the percentage overshoot in the unit-step response is 0.2 and the peak
time is 1 sec.
Solution: Note that
Performance of the First and Second Order
Systems
Since

Therefore, K and Kh can be obtained from


Effects of Zeros
We are considering the effect of a ‘zero’ on the unit-step response of a
second-order system:

R(s) C(s)

We only consider a real zero, that is, a is a real number.

Case I: a>0, i.e., the zero –a lies in the left-half complex plane.

We may have one of the following two situations:

1) G(s) possesses two real poles –b and –c (over-damped)


Effects of Zeros
Obviously, if a is close to either b or c, then a-b or a-c would be very
small so the system G(s) would be close to a first-order system.

Otherwise, if a is far away from both b or c, then we have:

That is, the effect of a is just like a scaling factor.

2) G(s) possesses a pair of complex poles (under-damped).


In this case, if |a| is far larger than |b|=|c|, then the effect of a
is still like a scaling factor as stated above.
Otherwise, the effect of a would be complicated and is shown in
the following figure.
Effects of Zeros

For this example, poles are at


Effects of Zeros
Case II: a<0, i.e., the zero lies in the right-half plane
A typical unit-step response is shown in the following figure. Note that
initially the response goes in the opposite direction, which is the typical
feature of a stable non-minimum phase system.

A system with either pole(s) or zero(s) in the right-half plane is called


a non-minimum phase system.
Performance of General Dynamic
Systems
In general, if a system is of higher order than 2, that is, includes three or more
Poles, or possesses more zeros , its performance evaluation could be
very complicated and needs to be conducted one by one. There is no systematic
conclusions reached about the performance like the situation of the first-order
and second-order systems. However, two special cases should be excluded:

1. If all poles are real, then a high order system can be treated as the
superposition of first-order systems.
2. In some special cases (these special cases actually cover quite a lot of
practical systems, and, hence, worth our attention), a higher order system
can be approximated as a second-order system by applying the so-called
dominant poles principle.

Dominant Poles Principle: If a system includes a pair of complex poles that are
close to the imaginary axis with all other poles (real or complex) locating relatively
far away from the imaginary axis, then the performance of this system will be
dominated by the pair of complex poles close to the imaginary axis.
Performance of General Dynamic
Systems
In this case, the high-order system will behave approximately as a
second-order system determined by the pair of dominant poles.

As an example, consider a three-poles system

Its unit-step response can be obtained as


Performance of General Dynamic
Three cases are shown in Systems
the figure:
1) Real pole is close to
complex poles
2) Real pole is far away
from complex poles
3) No finite real pole

In Case 1, dominant
poles principle cannot be
applied;

In Case 2, there are a


pair of dominant poles;

Case 1 and Case 2 are


compared with Case 3,
which is a second-order
system.
Tyre
Sensor disturbance Angle
- +
+ -
G1(s) G2(s) C(s) G3(s) G5(s) G6(s)
HW + - + - + -
Input Wheel Torque Controller Motor Gear Rack
Torque T(s) Shaft Sensor Road
Torque
G4(s)

Current Sensor

Transfer Function Model of A Electric Power Steering (EPS) System


Vehicle Speed
Hand Wheel

Controller
CAN
Torque
Sensor
Motor
Gear

Tyre Tyre

Rack

Schematic Diagram of A Electric Power Steering (EPS) System

You might also like